During Phase I, Flint Hills Scientific developed an algorithm for real time quantitative seizure detection which performs with sensitivity and specificity equal to expert visual analysis. Of even greater value is the capability of this algorithm to predict seizure onset by 13.6 seconds (mean), in its generic mode. Preliminary studies indicate that with automated individualized adaptation, prediction time can be increased to 180 seconds or longer. To the best of our knowledge no other system in existence has achieved this level of success. These results lay the ground for the fulfillment of "seizure prediction, early recognition and blockage of seizures," the number one AES research priority. The main goal of Phase II will be to advance, further refine, and validate this technology for implementation into a portable or implantable device with diagnostic, warning, and therapeutic capabilities. We are confident that this technology, by decreasing or eliminating unpredictability, will minimize the potentially devastating effect of seizures on quality of life while decreasing morbidity, the cost of health care, and the reliance on the welfare system. These unique advantages will ensure widespread acceptance of this technology by those directly and indirectly affected by epilepsy and by the health care system. PROPOSED COMMERCIAL APPLICATION: 1. Software package for real time seizure prediction, detection, localization, imaging, and quantitative analysis. 2. Software package for automated, selective noise reduction 3. Portable device for the automated early warning of impending seizures. 4. Portable or implantable devices for automated early therapeutic intervention.